This research explores the impact of interview mode on respondents' willingness to reveal illicit or undesirable behavior and mechanisms by which interview mode influences response tendencies. A field experiment was designed to control mode effects due to sampling and screening so that the impact of response anonymity (through use of self-administered questionnaires ([SAQs]) and social distance in the interviewer-respondent relationship (telephone vs. personal communication) could be tested. Respondents aged 18–45 were randomly as-signed to interview mode: telephone, face-to-face, or self-administered. Admission of illicit drug use and alcohol use was most likely in the personal mode with SAQs, slightly less likely in personal mode without SAQs, and least likely in the telephone mode. The magnitude of the mode differentials was larger for blacks than for whites, and larger among respondents who are more mistrustful of others. Results support the notion that response effects due to mode of interview derive, at least in part, from interview mode differences in ability to assuage respondents' confidentiality concerns. Greater social distance between interviewer and respondent in the telephone interview, compared with face-to-face communication, makes it more difficult for the researcher to make convincing confidentiality guarantees. The response anonymity provided by SAQs also appears to increase respondents' willingness to reveal sensitive behavior, especially among racial/ethnic minorities.